Editing Fusion power (section)

=== Machine learning ===
A [[deep reinforcement learning]] system has been used to control a [[tokamak]]-based reactor.<ref>{{cite journal |last1=Degrave |first1=J. |last2=Felici |first2=F. |date=2022 |title=Magnetic control of tokamak plasmas through deep reinforcement learning |journal=Nature |volume=602 |issue=7897 |pages=414–419 |doi=10.1038/s41586-021-04301-9 |pmid=35173339 |pmc=8850200 |bibcode=2022Natur.602..414D |url=https://doi.org/10.1038/s41586-021-04301-9 |access-date=February 5, 2025}}</ref> The system was able to manipulate the magnetic coils to manage the plasma. The system was able to continuously adjust to maintain appropriate behavior (more complex than step-based systems).{{cn|date=July 2024}} In 2014, Google began working with California-based fusion company [[TAE Technologies]] to control the [[Joint European Torus]] (JET) to predict plasma behavior.<ref>{{Cite magazine|last=Katwala|first=Amit|date=February 16, 2022|title=DeepMind Has Trained an AI to Control Nuclear Fusion|language=en-US|magazine=Wired|url=https://www.wired.com/story/deepmind-ai-nuclear-fusion/|access-date=February 17, 2022|issn=1059-1028}}</ref> [[DeepMind]] has also developed a control scheme with [[Tokamak à configuration variable|TCV]].<ref>{{cite magazine | url=https://www.wired.com/story/deepmind-ai-nuclear-fusion/ | title=DeepMind Has Trained an AI to Control Nuclear Fusion | magazine=Wired | last1=Katwala | first1=Amit }}</ref>